Large numbers of deterministic networks are in existence today. They are generally based either upon proprietary digital standards, upon enhancements to IEEE 802.3 Ethernet that are unlikely to become common, or upon overprovisioning of standard Ethernet bridged LANs. They are typically small and, if based on Ethernet, generally bridged rather than routed. They are characterized by the applications they enable—true real-time applications, that demand ultra-high packet delivery probability and reliable maximum latency guarantees. Most automobiles, planes, or trains built, today, incorporate one or more of these networks, as does many complex automated machinery on a production floor.
Classical packet network redundancy (e.g., bridging or routing protocols) are generally insufficient for achieving the low loss ratio required for these types of applications, as fault detection and recovery times are can be too slow by orders of magnitude. While overprovisioning, weighted queuing, and traffic shaping are useful for some genuine real-time applications, those techniques are not in themselves sufficient for at least two reasons. First, many critical applications require more bandwidth than these techniques can support (e.g., multiple uncompressed HD television streams). Second, weighting and shaping, at even the lower stream bandwidths, typically provide guarantees no better than 10−4 for congestion loss, and provide no absolute latency guarantee at all. Furthermore, these techniques are infamous both for requiring expert knowledge to match the controllable parameters to the application's needs, and for their inflexibility in the face of changing requirements.
Both operator-configured and application-requested path redundancy are generally needed for high reliability applications to overcome losses caused by equipment failure and by chance events such as cosmic rays. Redundancy alone is not sufficient, however, because congestion losses on the different paths are both non-negligible, and may be correlated. By eliminating the dominant cause of packet loss, congestion, the requirements of these applications can be met. Moreover, for time synchronization to the 1 μs level, some kind of hardware assist is likely required; software is generally too far from the physical media to support such accuracy.
As the size of the network grows, both geographically and logically, the requirements for latency and delivery probability relax. However, the requirements remain much higher than current networking techniques can support.
The presently disclosed systems and methods for flexibly reserving cycle-slots in ports of network nodes based on a binary schedule scheme are directed to overcoming one or more of the problems set forth above and/or other problems in the art.